HANGUPCAUSE = 1

[devplan1]

I noticed that the predictive dialer doesn't care about HANGUPCAUSE = 1 (which is unallocated number).

Shouldn't vicidial "auto-disposition" that call as "vacant number"?

Since the DIALSTATUS is too generic, the message played by asterisk to the agent when a vacant number is dialed doesn't give info whether the number is vacant or some other congestion cause.

So - how can this be implemented?

rgds, devplan1

[mflorell]

If that is always what that code meant I would have it do that, but after years of dialing in the USA I can tell you that 20-30% of calls over a PRI to numbers in the USA that report that code are not unallocated numbers.

Unfortunately this is the price we pay in the USA for cheap long distance and de-regulation of 20 years ago.

[marmu]

I am not in the US and would like to have more causes/statuses within vicidial. HANGUPCAUSE supports a lot more:

Code: Select all

/* Causes for disconnection (from Q.931) */
 #define AST_CAUSE_UNALLOCATED 1
 #define AST_CAUSE_NO_ROUTE_TRANSIT_NET 2
 #define AST_CAUSE_NO_ROUTE_DESTINATION 3
 #define AST_CAUSE_CHANNEL_UNACCEPTABLE 6
 #define AST_CAUSE_CALL_AWARDED_DELIVERED 7
 #define AST_CAUSE_NORMAL_CLEARING 16
 #define AST_CAUSE_USER_BUSY 17
 #define AST_CAUSE_NO_USER_RESPONSE 18
 #define AST_CAUSE_NO_ANSWER 19
 #define AST_CAUSE_CALL_REJECTED 21
 #define AST_CAUSE_NUMBER_CHANGED 22
 #define AST_CAUSE_DESTINATION_OUT_OF_ORDER 27
 #define AST_CAUSE_INVALID_NUMBER_FORMAT 28
 #define AST_CAUSE_FACILITY_REJECTED 29
 #define AST_CAUSE_RESPONSE_TO_STATUS_ENQUIRY 30
 #define AST_CAUSE_NORMAL_UNSPECIFIED 31
 #define AST_CAUSE_NORMAL_CIRCUIT_CONGESTION 34
 #define AST_CAUSE_NETWORK_OUT_OF_ORDER 38
 #define AST_CAUSE_NORMAL_TEMPORARY_FAILURE 41
 #define AST_CAUSE_SWITCH_CONGESTION 42
 #define AST_CAUSE_ACCESS_INFO_DISCARDED 43
 #define AST_CAUSE_REQUESTED_CHAN_UNAVAIL 44
 #define AST_CAUSE_PRE_EMPTED 45
 #define AST_CAUSE_FACILITY_NOT_SUBSCRIBED   50
 #define AST_CAUSE_OUTGOING_CALL_BARRED      52
 #define AST_CAUSE_INCOMING_CALL_BARRED      54
 #define AST_CAUSE_BEARERCAPABILITY_NOTAUTH 57
 #define AST_CAUSE_BEARERCAPABILITY_NOTAVAIL     58
 #define AST_CAUSE_BEARERCAPABILITY_NOTIMPL 65
 #define AST_CAUSE_CHAN_NOT_IMPLEMENTED      66
 #define AST_CAUSE_FACILITY_NOT_IMPLEMENTED      69
 #define AST_CAUSE_INVALID_CALL_REFERENCE 81
 #define AST_CAUSE_INCOMPATIBLE_DESTINATION 88
 #define AST_CAUSE_INVALID_MSG_UNSPECIFIED   95
 #define AST_CAUSE_MANDATORY_IE_MISSING 96
 #define AST_CAUSE_MESSAGE_TYPE_NONEXIST 97
 #define AST_CAUSE_WRONG_MESSAGE 98
 #define AST_CAUSE_IE_NONEXIST 99
 #define AST_CAUSE_INVALID_IE_CONTENTS 100
 #define AST_CAUSE_WRONG_CALL_STATE 101
 #define AST_CAUSE_RECOVERY_ON_TIMER_EXPIRE 102
 #define AST_CAUSE_MANDATORY_IE_LENGTH_ERROR 103
 #define AST_CAUSE_PROTOCOL_ERROR 111
 #define AST_CAUSE_INTERWORKING 127
 /* Special Asterisk aliases */
 #define AST_CAUSE_BUSY  AST_CAUSE_USER_BUSY
 #define AST_CAUSE_FAILURE  AST_CAUSE_NETWORK_OUT_OF_ORDER
 #define AST_CAUSE_NORMAL  AST_CAUSE_NORMAL_CLEARING
 #define AST_CAUSE_NOANSWER   AST_CAUSE_NO_ANSWER
 #define AST_CAUSE_CONGESTION   AST_CAUSE_NORMAL_CIRCUIT_CONGESTION
 #define AST_CAUSE_NOTDEFINED  0

Any hints on implementing that?

[mflorell]

Post the results of this MySQL query:

select count(*),hangup_cause from vicidial_carrier_log group by hangup_cause;

[williamconley]

mysql> select count(*),hangup_cause from vicidial_carrier_log group by hangup_cause;
+----------+--------------+
| count(*) | hangup_cause |
+----------+--------------+
| 125615 | 0 |
| 1261 | 1 |
| 346071 | 16 |
| 3976 | 17 |
| 53 | 18 |
| 136 | 19 |
| 446 | 21 |
| 1 | 27 |
| 7029 | 38 |
+----------+--------------+
9 rows in set (1.26 sec)


got tired of waiting

hey, 2.4 is cool. or was this also in 2.2.X?

[mflorell]

This is in 2.2.1, and actually the carrier log has been in there since early on in the 2.2.0 branch, but it was made required and active late in the 2.2.0 branch just before the release.


Here are the results for one of our clients that uses PRIs for the last two months:

mysql> select count(*),hangup_cause from vicidial_carrier_log group by hangup_cause;
+----------+--------------+
| count(*) | hangup_cause |
+----------+--------------+
| 6594825 | 0 |
| 24429 | 1 |
| 3 | 3 |
| 4288310 | 16 |
| 73537 | 17 |
| 874 | 18 |
| 6 | 19 |
| 25810 | 21 |
| 19149 | 28 |
| 598 | 31 |
| 50467 | 34 |
| 4 | 43 |
| 28 | 44 |
| 457 | 50 |
| 343 | 66 |
| 112 | 82 |
+----------+--------------+

And here is another client that uses all SIP for outbound for two months:

+----------+--------------+
| count(*) | hangup_cause |
+----------+--------------+
| 1253503 | 0 |
| 220 | 3 |
| 2118599 | 16 |
| 20380 | 17 |
| 109 | 19 |
| 3217 | 21 |
| 5 | 27 |
| 2327 | 28 |
| 29952 | 38 |
| 59 | 58 |
+----------+--------------+

Interesting that SIP actually sends less types of reason codes than PRI.

[williamconley]

i think i may have a client who is interested in "testing" these.

those that return with unallocated on one provider, call 1000 of them on another provider and see if ANY answer or come back with a response other than unallocated. i might expect 1 or 2 to get allocated in that time (coincidence), but certainly not a measurable percentage.

[mflorell]

last time we did that(a couple years ago) it was on different PRI providers, and 10-20% of unallocated numbers came back ring-no-answer or something else that was not unallocated when attempted multiple times within the next week(and that includes calls on the original carrier), and this was testing about 10,000 calls.

[williamconley]

sounds about right.